Claude status: is it down right now?
Our own check of claude.ai from Europe and the US, every 3 minutes, with recent history.
claude.ai refused our check. The server refused access. We do not know whether the service is up.
claude.ai refused our check.
The server refused access. We do not know whether the service is up.
Checked (just now) from Europe and the US, automatic every 3 minutes.
Run a website? Hear about it before your visitors do.
Visual Sentinel watches your own site and alerts you when a check fails or the page looks broken. Add your address and we'll check it now. No account needed.
This checks the response from claude.ai. It does not test playback, login or every Claude feature.
Claude status history
Our own checks from Europe and the US. Green means every check passed, amber means some checks failed, red means most checks in that period failed.
Outages our checks detected
None since Oct 13. A problem has to fail two checks in a row before we record it.
Watch your own site
This page is for their outage. Your own site can fail the same way. See its first check free, then try a paid plan for 7 days (card required).
Monitor my siteabout this service
About Claude
Claude is Anthropic's AI assistant, with a consumer product at claude.ai (chat, Projects, Artifacts, Computer Use) and a developer API at api.anthropic.com used by Cursor, Zed, Raycast, and thousands of LLM-powered apps. Claude Code, Anthropic's terminal/IDE agent, runs against the same API. Outages show up as chat responses failing mid-stream, Artifacts not rendering, Projects losing context between turns, or the API reporting that it is overloaded. Different model tiers (Opus, Sonnet, Haiku) can be affected independently, a Sonnet outage may leave Opus healthy and vice versa. Claude-powered agents hit rate-limit surges first during any capacity incident.
The official Claude status page is available as a reference. Official pages are useful, but they can lag behind real outages by minutes or hours.
outage patterns
Claude not working? What an outage looks like
AI assistants often fail in part rather than all at once. The site or app can open while replies stall, stop partway or end in an error, or while one model or feature is unavailable and the rest still work.
The check above requests claude.ai from Europe and the US every 3 minutes. It does not sign in, send a prompt or test your plan, so a passing check means the service responded, not that every model and feature works.
Common signs Claude is having issues
- Replies stop partway, never start, or end in an error message
- One model or feature fails while the others still work
- Sign-in loops, or the app keeps asking you to log in again
- Requests fail the same way on more than one device and network
Usually not a Claude outage
- A usage or message limit on your plan
- A company network, VPN or firewall blocking the service
- An expired session: sign out and sign in again
- A browser extension or ad blocker interfering with the page
What to do when Claude is degraded
Try again from a second device or network, and sign out and back in. If the check above or the official status page shows a problem, it is likely on Claude's side, so wait or switch to another tool for now. If both look fine, check your plan's limits and your network first.
monitor 24/7
Monitor Claude automatically.
Get independent uptime alerts when the public Claude endpoint fails a confirmed check. Solo and Starter check every 5 minutes; Business and Agency check every 60 seconds.
5-minute checks from Solo
Business and Agency support 60-second uptime checks.
Multi-region
EU & US checks. Catch regional outages.
Alerts where you work
Email and push on every plan, plus team and incident channels by plan.
free first check · 7-day trial, card required · from $15/month
why independent monitoring
Provider status references
and endpoint checks answer different questions.
Independent endpoint evidence
The regional check can show that a public endpoint is failing before or after a provider posts broader incident context. It does not validate every product subsystem.
Scheduled alerts
Monitor Claude on your plan interval and route confirmed endpoint failures through the notification channels available to your workspace.
Multi-region verification
EU and US results can reveal a regional difference in public-endpoint reachability. Confirm the exact workflow and provider region you use.
Historical data
Track Claude uptime over time. See trends, identify recurring issues, and make informed decisions about your dependency.
Frequently asked questions
Is Claude down right now?
See the latest check of claude.ai and its timestamp at the top of this page. If Claude is down, you can also check the official status page for extra context.
How do I check if Claude is having issues?
This page checks the public Claude endpoint from EU and US. You can also monitor that endpoint on a schedule and receive confirmed-failure alerts through the channels available on your plan. Test the exact provider feature you use for a complete diagnosis.
How can I get alerted when Claude goes down?
Add claude.ai to Visual Sentinel and its first check is free. To keep monitoring, start a 7-day trial (card required) or subscribe now: Solo and Starter check every 5 minutes, Business and Agency every 60 seconds, with email, push and chat alerts.
Where can I find Claude outage context?
The official Claude status page is available as a reference. Official pages are useful, but they can lag behind real outages by minutes or hours. Visual Sentinel checks the public Claude endpoint independently from EU and US. Paid Business and Agency plans support 60-second uptime checks.
Why does the Claude API say it is overloaded?
An overloaded response means Claude is temporarily limiting requests because it is at capacity. It is not an outage in the full sense; individual requests fail but the service is alive. Retry with exponential backoff, or fall back to a different model tier.
Does a Claude.ai outage affect Claude Code?
The consumer claude.ai web app and Claude Code are separate clients but call the same Anthropic API. A pure UI outage on claude.ai leaves Claude Code working; an API-side outage breaks both.
Often checked together with Claude
When an AI assistant stops answering, people usually try a second one. Each page below has its own check, run every 3 minutes from Europe and the US.